Enhancing Human Face Detection by Resampling Examples Through Manifolds

Abstract

Abstract-As a large-scale database of hundreds of thousands of face images collected from the Internet and digital cameras becomes available, how to utilize it to train a well-performed face detector is a quite challenging problem. In this paper, we propose a method to resample a representative training set from a collected large-scale database to train a robust human face detector. First, in a high-dimensional space, we estimate geodesic distances between pairs of face samples/examples inside the collected face set by isometric feature mapping (Isomap) and then subsample the face set. After that, we embed the face set to a low-dimensional manifold space and obtain the low-dimensional embedding. Subsequently, in the embedding, we interweave the face set based on the weights computed by locally linear embedding (LLE). Furthermore, we resample nonfaces by Isomap and LLE likewise. Using the resulting face and nonface samples, we train an AdaBoost-based face detector and run it on a large database to collect false alarms. We then use the false detections to train a one-class support vector machine (SVM). Combining the AdaBoost and one-class SVM-based face detector, we obtain a stronger detector. The experimental results on the MIT + CMU frontal face test set demonstrated that the proposed method significantly outperforms the other state-of-the-art methods.

Effective discretization of Gabor features for real-time face detection

“…For training, we use a database of 25000 cropped face images with size of 20 × 20 pixels, which includes faces with slight variations in pose angle and under different illumination conditions [7]. The negative samples are from a dataset of more than 25000 background images.…”

Face Transformation With Harmonic Models by the Finite-Volume Method With Delaunay Triangulation

“…21-23 with 74, 283, and 1312 nodes. 2 The harmonic transformation by the FVM using these three partitions is also chosen, and the CSIM only with μ = 0 is employed. The computed errors are listed in Table V. From Table V, the same convergence rates as (4.15) and (4.16) are observed.…”

“…Evidently, the numerical techniques in this paper may be applied to 3-D face data, reflecting recent trends in graphics and recognition work. Furthermore, the numerical transformations of face images in this paper can be used not only for pattern recognition but also for resampling [2], [25], imaging morphing [26], [27], and computer animation [20]. This paper is organized as follows: In Section II, the harmonic model, the FVM, and Delaunay triangulation are described.…”

Theoretical Analysis of LLE Based on Its Weighting Step

“…Here, a manifold is defined as a topological space that is locally equivalent to a Euclidean space. LLE was found to be useful in data visualization (Roweis and Saul 2000;Xu et al 2008) and in image processing applications such as image denoising (Shi, Shen, and Chen 2005) and human face detection (Chen et al 2007). It is also applied in different fields of science, such as chemistry (L'Heureux et al 2004), biology (Wang et al 2005), and astrophysics (Xu et al 2006).…”